INFRASTRUCTURE VICTORIA

Technical Paper
Lorraine Conway and David Prentice

How much do households

respond to electricity
prices?
Evidence from Australia
and abroad

Technical Paper No. 1/19

September 2019
 How much do households respond to electricity prices?
Evidence from Australia and abroad
Lorraine Conway and David Prentice12

Infrastructure Victoria Technical Paper No. 1/19

September 2019

Abstract

In this paper we review studies to understand how much households change their electricity

consumption when there is a price change. We are particularly focussed on finding results from

econometric studies that estimate elasticities of demand.

Many studies find residential households demonstrate responsiveness to price, with long term

and short run elasticities behaving as economic theory would suggest. For instance, the elasticities are

negative which means that as price increases, consumption decreases; long run elasticities are larger

than shorter run elasticities which indicates that households can respond over time through investment

in more energy efficient appliances; and very short run elasticities exist – while very short run

elasticities are small, household responsiveness seems to increase when paired with technology. Long

run elasticities range from -0.75 to -0.3 and short run elasticities range from -0.47 to -0.026. The major

gaps in research from the empirical economics literature are how low income and vulnerable Australian

households could be affected by price changes and how Australians respond to within-day variation in

prices.

1
We would like to thank Kath Rowley and Maria Wilton for detailed feedback. This paper has also benefitted from
conversations with colleagues within Infrastructure Victoria.
2
Corresponding author contact details: david.prentice@infrastructurevictoria.com.au
 1. Introduction

In Australia, the electricity market has undergone significant structural change over the last two

decades following competition reforms and the creation of the National Electricity Market. In addition,

electricity demand patterns have changed, with a significant impact on the need for infrastructure

investment. Although average consumption per head has been declining, demand is becoming more

and more peaked (AEMO, 2019). New maximum demand records continue to be set and the diffusion of

electric vehicles could further increase maximum demand. New investment in infrastructure is also

required due to changes on the supply side such as technological change in generation by renewable

sources and storage.

Key economic bodies in Australia, such as the Australian Competition and Consumer

Commission and the Productivity Commission, have recommended moving away from flat rate

electricity prices to more cost reflective pricing, such as time-of-day pricing, to support more efficient

use and investment in electricity infrastructure. There are several reasons why policy makers may have

been reluctant to take this step. Concerns that have been raised in a variety of forums include:

1. Effectiveness of differential pricing to manage demand by households — there is public debate

around whether households are responsive to substantial price differences.3

2. Impacts on vulnerable households — discussion tends to focus on potential negative impacts.

For example, charging higher prices when the demand for energy is greatest could place

financial burdens on consumers who are least able to change their usage. Alternatively,

financial constraints could prompt vulnerable households to cut consumption on particularly

3
See Ben-David (2018, page 9) and Finkel et al. (2017, pages 146-147). Ben-David states “studies have estimated
the long run elasticity of electricity demand lies between -0.08 and -0.15” such that the household has very little
opportunity to change its consumption in response to price. However, the studies are not identified. Our review
documents studies that household responsiveness is much larger than -0.15. Finkel et al. (2017) provides evidence
of the effectiveness of responding to financial incentives, but also notes that overseas evidence suggests the
responsiveness of households to price signals is difficult to sustain.

1
 hot days.4 However, for those low income households who can respond, cost reflective pricing

may create opportunities to make savings leaving them better off than under flat rates.5

3. A perception that cost reflective pricing will lead to, in general, paying more for electricity.6

This paper aims to fill the evidence gap for Australia on the first of these concerns through

examining the existing economics literature on the responsiveness of households to changes in

electricity prices.7 We identify and present existing estimates of demand elasticities and complementary

evidence on how households respond to prices i.e. empirical evidence of the percentage change in the

quantity of electricity demanded in response to a percentage change in the price of electricity (also

known as the own-price elasticity of demand). We focus on both the long-run and very short run

elasticities of demand. We also consider the influence of new technology on demand responsiveness.

To better understand the second set of concerns, we also examined the literature for how price

responsiveness may differ by cohort — especially for lower income and other vulnerable households.

We also briefly consider some work that analyse, under no behaviour change, whether different types

of households spend more or less under a form of cost reflective pricing.

Our first set of conclusions is that there is strong evidence that Australian households respond

to electricity prices in the long run with long run own-price elasticities ranging from -0.75 to -0.3. This

means that, in response to a one per cent increase in the price of electricity demand may fall by

between 0.3 and 0.75 per cent. The degree of price responsiveness of demand for electricity is

comparable, at the upper bound, -0.75, to that of several other products commonly seen as necessities,

such as bread and milk. In general, the demand for electricity is more price responsive than the demand

for other necessities, like petrol and water.

The second set of conclusions is that international evidence suggests households will respond to

incentives to shift consumption within a day. The associated elasticities vary from -0.026 to -0.47.

4
See Nicolls et al. (2017) which highlights the concerns that a set of vulnerable consumers have about a proposed
cost reflective pricing system, as well as evidence around the consequences of financial constraints for electricity
consumption.
5
See Stenner et al. (2015) which assessed consumers’ likely uptake of cost reflective pricing.
6
This sort of concern is raised by the, particularly vulnerable, participants in the study by Nicolls et al. (2017)
7
Ausgrid (2015) does a brief summary of some Australian and international work.

2
Demand responsiveness is further increased by complementing price signals with technology, which

makes it easier for households to find out about price differences and respond to them.

The third set of conclusions is that there are two significant research gaps. First, there is a

shortage of Australian evidence on how households will respond to incentives to shift their

consumption within a day. Second, further evidence is needed to better understand how different types

of low income and vulnerable households actually respond to cost reflective pricing, and the

effectiveness of different policies to mitigate any risks.

These findings have three policy implications. First, as household demand for electricity is

responsive to price changes, policies that affect electricity prices will, holding all other variables

constant, affect household consumption. Second, any policies that affect electricity prices should,

where possible, be combined with technological and regulatory changes that improve consumer

information about prices and the ability to respond to them. Finally, future research should prioritise

finding out more on how Australian households respond to more dynamic, time-varying electricity

pricing with a particular focus on low income and vulnerable households. With its extensive network of

smart meters (97% of retail customers), Victoria is particularly well placed to gather more evidence

either through examination of existing data or through trials.8 The lessons would also be relevant for

other states.

In the next section, Section 2, we discuss how prices are an essential feature of energy policy.

This is followed by a discussion of findings in the literature relating to long run elasticities in Section 3

and short run elasticities in Section 4. Both Section 3 and 4 draw on Australian and international studies

of household electricity demand. Section 5 examines literature on the responsiveness of different

households, particularly lower income households, followed by a short discussion and concluding

comments in Section 6.

8
Victorian Department of Economic Development, Jobs, Transport and Resources, Smart meters: End of rollout,
fact sheet, March 2015.

3
 2. Prices as a part of energy policy

Most households in Australia pay for their electricity using relatively flat price schedules with

limited on and off-peak differential in prices. However, Infrastructure Victoria and leading economic

bodies in Australia have argued for more variable price schedules to improve the efficiency of use of

electricity infrastructure and in turn, investment in infrastructure. In this section we review these

arguments and the type of pricing structure recommended by these bodies: cost reflective pricing.

2.1 Background on cost reflective pricing

The price of electricity is linked to infrastructure costs as these must ultimately be recovered,

albeit subject to regulation, from consumers. As most users are currently unable to store electricity

infrastructure must be built to a capacity that can provide electricity when demand peaks on a few hot

days each year — otherwise there can be load shedding. The greater the level of peak demand, the

greater the cost of infrastructure required to meet it, and the higher the price of electricity must be to

recover these costs. Moving away from flat rate electricity pricing could moderate peak demands

resulting in better use of the infrastructure and therefore less need to build additional infrastructure.

Keeping other variables constant, this will result in lower electricity prices.9 We now explore the

potential role of cost reflective pricing in achieving these outcomes in more detail.

Cost reflective pricing refers to a pricing system that more closely ties the price paid for

electricity to the cost of supply, including the required infrastructure. Cost reflective pricing includes a

wide range of pricing structures, also known as tariff structures. The structures can include time-of-use

pricing where blocks of time are charged with different prices, and critical peak pricing where a

significantly high price is charged for when system demand is highest (usually reserved for a certain

number of days a year). There are also variants on these types of pricing such as peak time rebates,

where households are given rebates for electricity that is not consumed, and variable peak pricing,

9
In its 2017 inquiry, the ACCC recognised that tariff reform has the potential to both reduce the total costs of the
network and make charges more equitable. It also stated that ‘demand response’ programs, such as rebates to
short term reductions in usage, are likely to provide a more immediate impact on managing network demand;
however, they do not address the issue of equity in the allocation of network cost.

4
where the price at the time of the peak demand depends on the level of overall electricity demand

during the peak.

Demand varies systematically with time and cost reflective pricing structures can reflect the

different underlying costs for the network at different times. This can guide more efficient use of

electricity and better use of existing infrastructure. It can provide consumers with signals to reduce their

consumption in periods of higher peak network use. In addition, it means more of the infrastructure

costs are born by those that make use of the peak infrastructure (and those who don’t, don’t have to

bear the costs). Cost reflective prices also provide clearer signals to firms of when there is sufficient

demand to justify further investment in the network. Borenstein (2013, page 127) summarises the

appeal of cost reflective pricing to energy economists:

“Economists who study electricity markets are virtually unanimous in arguing that time-varying

retail pricing for electricity would improve the efficiency of electricity systems and would lower

the overall cost of meeting electricity demand.”

2.2 Cost reflective pricing in Australia

In Australia, there are institutions and laws to enable the adoption of cost reflective pricing. For

instance, it is an embedded pricing principle for electricity distribution networks, set by national

agreement through the Council of Australian Governments (COAG) Energy Council in 2013 and put into

effect by the Australian Energy Market Commission through a rule determination in 2014. And, in

Victoria, at least, the smart metering technology required to implement cost reflective pricing is also in

place.10

Nevertheless, there has not been wide spread adoption of cost reflective pricing by households.

Electricity companies can offer cost reflective pricing, yet the tariffs for the majority of small customers,

such as households, are largely flat tariffs regardless of when that electricity is used (ACCC, 2018). One

likely reason for this is that the approach to implementation in Victoria is an opt-in basis. This means

customers remain on their existing tariff structure unless they actively choose a new tariff structure. The

10
Victorian Department of Economic Development, Jobs, Transport and Resources, Smart meters: End of rollout,
fact sheet, March 2015.

5
Victorian Government’s preference for an opt-in model is to provide households with choice about

what pricing arrangements are suitable for their circumstances.11 However, there are limited incentives

for customers to switch to cost reflective pricing, limited incentives for retailers to translate it into value

for customers and limited incentives for governments to promote it (ACCC, 2018). Under a flat tariff

regime, some households are currently subsidised by others and have little incentive to change to a

more cost reflective arrangement.12 Opt-in generally results in a lower uptake compared with opt-out,

which we discuss in more detail in Section 4.

Key economic bodies in Australia have argued for cost reflective pricing based on traditional

static efficiency grounds. In 2016, Infrastructure Victoria released Victoria’s 30-year infrastructure

strategy and made recommendations to the state government across all infrastructure sectors including

energy. A key energy recommendation was to facilitate a reduction in consumption by mandating cost

reflective pricing. Infrastructure Victoria argued that a more efficient pricing system would result in

better use of existing infrastructure and efficient investment in the future. In 2018, the ACCC also made

a recommendation for mandatory cost reflective pricing as part of its inquiry into the retail electricity

market, accompanied by supporting recommendations to address transitional issues.13

2.3 Changing electricity market could be better supported through cost reflective pricing

The case for moving towards cost reflective pricing is strengthened by the current and potential

pressures for new investment in generation and the network in general as a result of two sets of shocks.

On the demand side, the uptake of electric vehicles will substantially increase the demand for

electricity, possibly further increasing the peak demands on the system. Infrastructure Victoria (2018)

argues that a more efficient outcome than just investing to meet this demand would be achieved by

11
See quote from the Victorian Minister for Energy in the Choice article, Will Victorians opt in to a new flexible
electricity pricing model?¸2016, accessed 22/11/2018 at www.choice.com.au/shopping/shopping-for-
services/utilities/articles/flexible-electricity-pricing-opt-in-victoria. The new Victorian and national default offers
are not cost reflective either (See: https://www.esc.vic.gov.au/electricity-and-gas/prices-tariffs-and-
benchmarks/victorian-default-offer and https://www.aer.gov.au/retail-markets/retail-guidelines-reviews/retail-
electricity-prices-review-determination-of-default-market-offer-prices for the national ones. (accessed 9/9/2019)
12
Wood and Carter (2014) provide a case of a Victorian consumer who purchases an air conditioner with a
capacity rating of five kilowatts. The authors estimate that the air conditioner could add between $1,200 and
$1,550 to the cost of the network, but its owner would only pay an extra $53.40 a year in network charges. The
rest is recovered from all users.
13
See Appendix 1 for a summary of recommendations by Infrastructure Victoria and the ACCC on cost reflective
pricing for electricity.

6
adopting mechanisms to shift demand away from the peaks to during the day. Cost reflective pricing is

one of these mechanisms.

Similarly, on the supply side, substantial new investment will result from the combination of

emission reduction goals, the exit of coal-fired generators, falling costs of renewables and batteries and

the geographic dispersion of generation away from the La Trobe Valley. Cost reflective pricing,

dampening peak demand, reduces the investment required by encouraging the most efficient use of the

new energy infrastructure. Less investment means less costs to be recovered from households through

energy bills.

2.4 The empirical question – do households respond to price changes?

The effectiveness of cost reflective pricing in reducing peak demand and delaying additional

infrastructure investment depends on the extent to which prices affect household decisions about the

quantity and timing of electricity consumption. As mentioned earlier, some commentators have

suggested that households do not respond very much to price changes (Ben-David, 2018; Finkel et al.,

2017). This is an empirical question. The main way the household responsiveness to prices is measured

by economists is by estimating the price elasticity of demand, which returns the general proportional

change in consumption to a one per cent change in price. This provides a rate of change at a more micro

level which can then be applied in economic modelling.

2.5 Why are elasticities important for policy makers?

In this paper, we focus on finding elasticities. The price elasticity of demand is important in

understanding the potential effectiveness of policies that affect the price of electricity and the

associated impacts.14 If the price elasticity of demand for electricity is not zero, then policies affecting

prices can affect behaviour and the larger the elasticity the greater the effect. How these elasticities

differ between the very short run, the short run and the long run illuminates the time path of impacts

following the introduction of such a policy. Finally, if we can see that different household types have a

different response to price based upon their specific attributes (i.e. income), this enrichens our

14
See Appendix 2 for background on demand curves, elasticities and their broad determinants.

7
understanding of the impacts and the necessity for compensation or adjustment assistance for different

cohorts.

Elasticities are also particularly beneficial for better understanding potential transitional impacts

of reform and for creating targeted responses. As an example of how knowledge of elasticities can

inform policy, Grafton and Ward (2008) demonstrate that households have been found to be responsive

to prices for water, and that the welfare costs from volumetric pricing of water to regulate water

demand is less than the welfare cost from mandatory water restrictions (Grafton and Ward, 2008).

For electricity, if low income households have inelastic demands, until the savings on

infrastructure costs are passed on, their expenditure on electricity could go up. Alternatively, if low

income cohorts have relatively elastic demands, cost reflective pricing may result in significant cuts in

consumption, particularly on hot days, until prices fall. This could have serious health consequences for,

for example, older low income Australians. Targeted responses could include some form of income

supplement similar to that which was provided when the carbon price was implemented but this would

need to be reviewed to ensure it largely dealt with potential risks. As households adjust their appliances

and make other changes, the size of the supplement required may decline. Another approach could be

to use regulation.

There are two limitations of the literature we are about to review. First, because of the very

limited diffusion of cost reflective pricing across Australia, the elasticities we will review are, unless

noted otherwise, for households facing flat rate prices. Second, the quantity data typically refers not to

demand at a point in time but consumption over a period of time. Cost reflective prices would be

expected to provide stronger incentives to moderate demand than consumption but the two are

correlated. And the elasticities are still informative as to whether, even when households are not facing

pointed price incentives, households will respond to price changes. Hence, in the next two sections we

review the evidence on the long run and short run price elasticities of demand for electricity, followed

by Section 5 which considers the very limited empirical evidence on differences in elasticities by cohort.

8
 3. Long run price elasticities of demand for residential electricity

Long run price elasticities of demand capture the responsiveness of households after they have

had as long as possible to respond to price changes. For example, they can also change their electrical

appliances, household insulation or make other substantial changes in response to a price change. They

are typically estimated over a substantial period and can produce an annual elasticity or even multi-year

elasticities. A set of estimates of long run elasticities of largely residential demand is presented in Table

1.

3.1 Long run elasticities of demand for electricity in Australia

The empirical evidence demonstrates that in the long run the demand for electricity will fall as

the price of electricity rises. The literature summarised in Table 1 suggests the long run elasticity for

electricity demand is somewhere between -0.75 to -0.3. Although the demand for electricity is inelastic

it is substantially different from zero so consumers will respond substantially to changes in prices.

Specifically, a recent analysis of Australia-wide residential demand for over 40 years (1970 to

2011) (Rai et al., 2014) finds a long-run price elasticity of -0.748 and a short run elasticity of -0.447.

Using a shorter, slightly older, time series (1969 to 2000), Narayan and Smyth (2005) find more

conservative annual estimates of -0.541 in the long run and -0.263 in the short run. Using a structural

break test, Rai et al. (2014) found that the price elasticity fell from -0.5 during 1970 to 1982 to -0.3

during 1982 to 2011. This suggests that there is some evidence that the long run elasticity has changed

over time and people have become less responsive in the long run in more recent decades.

There is more limited work at the state level in Australia which typically analyses total (not just

residential) consumption — though it is not clear which way this biases the results. Fan and Hyndman

(2011) used half hourly data for South Australia from 1997 to 2008 to develop annual own price

elasticities of total demand (excluding demand associated with mining) for South Australia and found

values of between -0.428 to -0.363. Similar values are found nationally by Doojav and Kalirajan (2019).

We note a survey paper by the National Institute of Economic and Industry Research reports elasticities

of between -0.53 and -0.23 for Victoria and a long run own price elasticity of demand between -0.5 to -

9
0.2 for the National Electricity Market15 (NIEIR, 2006). However, the methodology used by the NIEIR is

not specifically described so our preferred estimates do not reflect this work.

Those studies which also estimate short run elasticities of demand find these are smaller than

the long run elasticities. This is consistent with economic theory and possibly results from capital

adaptation or learning.

The long run elasticities for electricity are comparable to those, for Australia, for other products

seen by some consumers as necessities, as summarised in Table 2. The demand for all of these products

is inelastic— demand responds less than proportionally to price changes — but is still responsive to

prices. The long run demand for electricity is a lot more elastic than the demand for petrol or water. The

long run elasticity of demand for electricity is most comparable to that for cigarettes. The upper bound

estimates for electricity own-price elasticities are comparable to those for milk and bread.

In addition, the results for long run elasticities in Australia for electricity are more elastic than

those internationally as estimated in a meta-analysis (Labandeira et al., 2017).

15
Created in 1998, the National Electricity Market combines the wholesale electricity markets of Victoria, New
South Wales, Queensland, Australian Capital Territory, South Australia and Tasmania.

10
Table 1: Summary of long run price elasticities of demand for residential electricity
Authors Method Time period/market Results

Rai et al. (2014) Error-correction model Australia-wide and Long run model
Australian states, 1970 Australia
to 2011. -0.748
Australian structural
break tests
Fallen from -0.5 to -0.3
over time
Short run model
Australia
Own price: -0.447
Narayan and Smyth Time series using Australia, 1969 to 2000 Own price
bounds testing Long run: -0.541
(2005) approach Short run: -0.263

National Institute of Review of own work Australia-wide and Australia

and review of overseas Australian states, 1980 Residential: -0.25
Economic and Industry and Australian to 1995. Commercial: -0.35
literature Industrial: -0.38
Research (2007) National Electricity
Market
-0.50 to -0.20 (mean -
0.35)
Victoria
-0.53 to -0.23 (mean -
0.38)
Deryugina et al. Natural experiment, Illinois (US), 2007 to Own price: -0.09 in
flexible difference-in- 2014. first six months, up to
(forthcoming) differences matching -0.27 after two years
approach
Combined residential and industry
Doojav and Kalirajan Autoregressive- Australia-wide and Australia
distributed lag model Australian States, 1999 Long run: -0.38
(2019) to 2013. Short run: -0.30
Victoria
Long run: -0.19
Short run: -0.12
(statistically significant
at the at 10% level)
Fan and Hyndman Linear demand model Annual, South Ranges from -0.428 to-
Australia, 1997 to 2008 0.363
(2011)

11
Table 2: Other estimates of demand elasticities for electricity, gas and other necessities

Paper Product Elasticity

Australia
Breunig and Gisz (2009) Petrol Short run: -0.13
Long run: -0.20
Grafton and Ward (2008) Water -0.17

Pham and Prentice (2013) Cigarettes -0.4 (Sydney) to -0.26 (Brisbane)

Hasan and Sinning (2018) Various foods Milk: -0.732

Bread: -0.781
Baby food: -0.136
International
Labandeira et al. (2017) Electricity (Short run): -0.126
Electricity (Long run): -0.365
Natural Gas (Short run): -0.180
Natural Gas (Long run): -0.684

3.2 Long run elasticities of demand for households?

The Australian long run elasticities of demand use data on households aggregated to the state

or national level. Unfortunately we did not find long run elasticity estimates for Australia using data

reported at the household level. While this informs us about households as a group respond it doesn’t

necessarily inform about the typical individual household response.

Looking overseas, Deryugina et al. (forthcoming) provides results that are closest to estimates

of elasticities for households. They examined the Illinois Municipal Aggregation program, which was a

program that allowed communities in Illinois to aggregate their demand for the municipality to

negotiate for a deal with an electricity supplier. The authors found that demand increased when there

was an expectation that price would decrease. In the first six months following a community decision to

aggregate, price elasticity of demand was estimated at -0.09, which increased up to -0.27 after two

years.

The way the program was implemented created a de facto natural experiment, which means

that Deryugina et al. are confident these represent causal effects. The estimates are broadly similar to

the long run estimates using annual data, albeit lower.

12
3.3 What this means

We conclude, based on empirical evidence for Australia, that in the long run the household

demand for electricity does fall in a material way as price increases. This implies that, in the long run,

reforms that increased the price of electricity would reduce the household demand for electricity.

The work of Deryugina et al. (forthcoming) suggests a promising direction for further research

to be undertaken for Australia. Following households or even small communities and observing their

demand by month (rather than by year) over several years offers the promise of learning more about

the responsiveness of demand as households adjust to the new price policy over time. Working with

higher frequency data enables the analysis to be confined to a period where there is a stable

relationship between variables. In addition, it also offers greater surety that we are estimating a causal

impact rather than mere correlation.

In the next section we analyse the evidence on the short run elasticity of demand for electricity

in Australia.

4 Short run price elasticities of demand for residential electricity

Short run elasticities of demand for electricity focus on how responsive households are within

much shorter time frames. In the short run, determinants of electricity demand like insulation and the

costly appliances in the house remain unchanged so the main way the consumer can respond is by

altering how they use these appliances and electricity in general. Hence these results are more relevant

for assessing the potential impact of cost reflective pricing. The available estimates are reviewed in

Table 3.

The first set of studies focus on household responses to the price schedules in the monthly bills

with different incentives to moderate consumption. The elasticities from these studies are much lower

than the long run elasticities — in one case completely inelastic. We explore the reasons for these

results below.

In contrast, studies that focus on daily or within daily responses by households to prices tend to

find that households reduce consumption when the price is higher. Like in the long run, demand varies

less than proportionally with price changes i.e. it is inelastic. Furthermore, several studies suggest that

13
the responsiveness of households is greater when price reform is accompanied by new technology that

makes it easier for households to observe and respond to price changes. Furthermore, some of the

studies reviews suggest that more households will take up cost reflective pricing packages if they are

made to be opt-out rather than opt-in.

After briefly reviewing the studies which focus on monthly consumption, we then concentrate

on those studies that estimate the responsiveness of household demand to price changes that attempt

to shift consumption within a day, on days with peak demands or are provided with short notice. These

studies typically find stronger results than those using monthly data. However, as nearly all these

studies have been done overseas, we highlight a major research gap in that there are almost no studies

of this type for Victoria or Australia.

4.1 Studies using monthly electricity consumption

There are two studies that considered the effect of different types of prices on household monthly

consumption. Ito (2014) finds monthly consumption does respond to average price changes with a

statistically significant elasticity of -0.054. However, the same study does not find evidence of

households responding when per-unit prices increase as monthly consumption exceeds a threshold. In

addition, the average elasticity of demand with respect to average price increased to -0.082 when a

monthly lag was considered, which could reflect some adaptive learning. Ito suggests two reasons for

the lack of response to marginal incentives. First, households may find the more complex pricing

systems too hard to respond to so they just respond to the average prices which they can estimate.

Secondly, households do not observe their consumption during the month and so cannot alter their

behaviour around thresholds. Hence, they respond to average prices instead.

14
Table 3: Summary of short run price elasticities of demand for residential electricity

Authors Method Time Results

period/market
Monthly and six monthly elasticities
Byrne et al. (2019) Experiment Victoria (Australia), Monthly own price: -0.003
October to (statistically insignificant)
November 2015.
Ito (2014) Natural Orange County, Monthly response to average price:
experiment, California (US), -0.054 (-0.082 with a lag)
panel data with 1999 to 2007 Monthly response to marginal
fixed effects price: 0.0002 (statistically
insignificant) (0.006 with a lag)
Deryugina et al. Natural Illinois (US), 2007 Own price: -0.09 in first six months,
(forthcoming) experiment, to 2014. up to -0.27 after two years
flexible
difference-in-
differences
matching
approach
Combined household and business
Kim et al. (2015) Natural Ontario (Canada), On-peak: -0.10 (not statistically
experiment May 2006 to significant)
October 2014 Mid-peak: -0.25 (statistically
significant)
Off-peak: -0.06 (statistically
significant p<0.10)
Daily or within-day elasticities
Allcott (2011) Field experiment, Chicago (US), 2003 Day-ahead notice of hourly prices,
randomised to 2006 overall reduced form demand
control elasticity: -0.1
Gillan (2018) Natural field California (US), Bounds of own price: -0.050 to -
experiment, 2017 0.470
utility model
Faruqui et al. Meta-analysis of Global A 10% increase in the price ratio
(2017) dynamic pricing would result in a 6.5% decrease in
trials using linear peak usage. With technology, this
regression to decrease was 11.1%.
estimate effects
of peak/off-peak
ratio
Faruqui et al. Analysis of Connecticut (US), Daily own price: -0.026
(2014) experiment 2009 Time substitution response to price:
results on daily -0.081 to -0.047price only; -0.128 to
and within-day -0.047 price and technology
consumption.

15
 In Australia, there was an experiment with residential customers in Victoria undertaken over

two months in 2015 analysed in the unpublished study of Byrne et al. (2019). Customers were provided

temporary discounts on their per unit consumption and on the fixed infrastructure component of the

bill. The study found that residential electricity demand is largely unresponsive to large reductions in

both marginal and average prices, with an own price elasticity of demand estimate of -0.003

(statistically insignificant) (Byrne et al., 2019). The authors note that this could reflect that the pricing

did not provide the sharp incentives to reallocate consumption like that done by time-of-use or critical

peak pricing. In addition, the discounts were not long enough to encourage substantial behaviour

change or the replacement of appliances. This may be one of the reasons for the different effects to the

Ito (2014) study which was not a trial but based on the ongoing prices charged by utilities.

A third, unpublished, study analysed changes in average aggregate (households, small business,

commercial and industrial) monthly consumption due to changes in average monthly price under time-

of-use pricing in Ontario (Kim et al., 2015). The study found on-peak demand had an elasticity of -0.10

(not statistically significant); mid-peak demand had a statistically significant elasticity of -0.25; and off-

peak had an elasticity of -0.06 (significant at the 10 per cent significance level). A potential problem with

this study is that industrial and commercial consumption included in the dependant variable is not

subject to time-of-use pricing.

There is a tension between the results of Byrne et al. (2019) and Ito (2014) and those from the

studies we review next in the next sections which cover studies of responsiveness by day or within a

day. These studies tend to find that consumers do respond to price incentives. We discuss in section 4.5

how to reconcile the different outcomes using monthly and daily and within-daily data.

4.2 Daily or within-day elasticities

There is no published research reporting elasticities for how Australian households respond to

prices by day or within a day. When consumers face prices that vary within a day or across days these

are more immediate price signals that households can respond to. The response to within day price

variation is of most interest to understand responsiveness to pricing for a peak demand event. Hence

16
the lack of studies for Australia is a significant research gap. However, there has been considerable

international research which we will now review.

An example of this type of research is Faruqui et al. (2014) which analysed a 2009 pilot in

Connecticut, USA, of 2,200 residential and small commercial and industrial customers and assessed the

difference of responses in hourly consumption to time-of-use pricing, critical peak pricing and peak time

rebates. Households were found to shift their consumption from peak to off-peak periods when there is

a change in the ratio of peak to off-peak prices with elasticities ranging from -0.047 for time-of-use

pricing, to -0.052 for the peak time rebate, to -0.081 for those on critical peak pricing (i.e. households

were almost twice as responsive to critical peak pricing) (Faruqui et al., 2014).16 When critical peak

pricing was combined with technology that automatically adjusts the air conditioning unit or

thermostat, the elasticity increased a further 50 per cent to -0.128. When analysing the effects on daily

consumption, the authors found that time-of-use pricing had no statistically significant effect, but that

critical peak pricing and peak time rebate both reduced daily consumption (daily elasticities of -0.026

with price, and with price and technology).17

The effect of combining very short run price signals (rather than those through the monthly or

quarterly bill) with technology to make it easier for households to learn and respond to the signals has

been of considerable interest in this literature. Technology could increase household price elasticity of

demand by allowing households to more easily observe prices and consumption, and adjust appliances

in the more immediate term (Allcott, 2011).

In general, the studies of how households respond to prices in the very short run find that

consumers respond to near time prices, and respond substantially more so when paired with enabling

technology that can automate household appliances, or otherwise assist households, to respond to near

time prices. Analysis of a database which contains evidence from various dynamic pricing pilots found

that consumers do respond to very short term price incentives (Faruqui et al., 2017). In addition, the

16
That is, a one per cent increase in the peak to off-peak price ratio implies that the usage ratio decreased by 4.7
per cent for those on time-of-use pricing and 8.1 per cent for those on critical peak pricing.
17
That is, a one per cent increase in price implies a decrease in overall consumption of 2.6 per cent.

17
 Figure 1: Summary of North American pricing pilots

Source: Newsham and Bowker (2010)

18
average effect on the peak across the different types of cost reflective pricing with enabling

technology was found to be greater than then average effect on the peak than those without

enabling technology.

An earlier review of North American studies also found that technology to automate

responses to price signals appears to increase responsiveness (Newsham and Bowker, 2010). Figure

1 on the previous page presents these results with darker shaded bars representing the demand

response to pricing with technology. Critical peak pricing, in particular, was found to be generally

very effective at reducing peak loads in the earlier review of North American pilots (Newsham and

Bowker, 2010).

4.3 Other studies of how consumers respond to within-day price variation

There are also several other studies that do not estimate elasticities but provide estimates of

the impacts of types of within day pricing as well as having other interesting features.

First, the different effects of dynamic peak pricing and peak rebates were analysed, using

the Smart Grid Smart City dataset from Sydney and surrounds in 2013. Households who signed up

for network dynamic peak pricing or a peak rebate changed their behaviour after subscribing to the

products. While an elasticity was not reported, the study found that the response was greater under

the dynamic peak rebate (usage reduction of 12.04%) than the dynamic peak pricing (usage

reduction of 8.71%) (Motagh et al., 2015).

Secondly, Allcott (2011) evaluated a program that exposed residential customers to hourly

prices based on a day-ahead wholesale prices— this is a version of real time pricing but with notice

of the price for each hour provided one day ahead. Key findings from the evaluation included that:

 program households are responsive — if average price was higher by one cent/kWh, the

demand reduction is around 17.4 watts in summer and 21.8 watts in non-summer.

19
  energy management and information technology, such as the Pricelight18, can significantly

increase a household’s responsiveness.

 program households reported that they responded to real time pricing through conserving

energy, rather than shifting load from high to low price times.19

Allcott (2011) also examined the original treatment group after one year on the program and

found that they consumed six to eight watts less on average when the price was higher by one per

cent – so the treatment group increased their responsiveness further by about 30%.

Thirdly, in a recent study based on a field experiment in California, Gillan (2018) found

households with automation technology were more responsive than those without, but also found

that households with notification of a pricing event still exhibited some responsiveness to price. In

this experiment, households were provided with less than an hour notice of a pricing event and

received an offer to reduce consumption for a financial incentive. Some households received

notifications, some received notice along with an environmental message (moral suasion) and some

received notifications paired with an incentive to purchase technology that could automate

responses to price events. The author found households which received a pricing notice, with and

without automation technology, responded to a price increase by reducing consumption, but had

limited response to the price level itself (i.e. consumption change is the same, regardless of the size

of the price increase) — which indicated that people use heuristics20 to make decisions. Households

that were notified of a pricing event reduced consumption by 12.4% on average. Households with

the automation technology reduced their consumption more than households without the

18
Pricelight is a device that changes colour when there is a change in price, providing a visual prompt for a
household to respond to. Farqui et al. (2014) reports mixed results for a similar technology, the Energy Orb.
19
This was found through a small survey of households in the program. When asked about what changes
participants thought they made after entering the program, respondents largely reported energy efficiency
measures rather than substitution. However, the author notes that within day substitution (i.e. using more in
the off-peak times compared to the peak, such as pre-cooling before the peak) could not be measured in the
evaluation.
20
Heuristics are mental shortcuts that ease the cognitive load of making a decision. For instance, using rules of
thumb or an educated guess are heuristic methods. In this particular study, it was theorised that households
could be using expected price change instead of the actual price change, such that demand is insensitive to the
magnitude of the price change (Gillan, 2018)

20
automation technology. Comparing consumption during the price event to consumption two hours

earlier for non-automated households, non-automated households had reduced their consumption

by 7.9% whereas the automated households had reduced their consumption by 23.4%. In addition,

the author found the response was precise as participants returned to normal consumption levels

post event.

Finally, in Victoria, Australia, the electricity distribution company, Jemena, in partnership

with the Department of Environment, Land, Water and Planning, Victorian Government, ran a pilot

over the summer of 2017/18 to test two forms of peak time rebates, made available at six, roughly

peak, trial times during the trial period. Households were given 48 hours notice of when the trials

were to be held and those that met a pre-set electricity target reduction received rebates either for

themselves or a local school or community organisation (Jemena, 2019). Households chose whether

to participate in the trial as a whole and whether to participate in each trial. On the two hottest days

average consumption reductions, compared with a control group, were between 26 and 35 per cent

of individual base peak consumption. Average participation rates in each trial were between 43 and

53 per cent. There were greater average participation rates but lower percentage reductions for the

group receiving the personal rewards compared with the group for whom the school/community

organisation received the rewards.

4.4 Opt-in, opt-out regimes

Another area examined in the literature is the different effects of using opt-in, opt-out or

mandatory assignment for cost reflective tariff structures. Generally, opt-out and mandatory

assignment are more effective than opt-in because more people take up the cost reflective pricing.

From a review of studies across six countries, Nicholson et al. (2018) found that with opt-out

arrangements, between 57% and 100% of customers remained on cost reflective pricing, whereas

opt-in models have a take up of cost reflective pricing arrangement of between 1% and 43% of

customers. Providing upfront financial incentives for signing up to time of use pricing was found to

21
have a strong statistically positive effect on uptake, but uptake is not suggested to be related to the

presence of automation technology.21

On whether the type of regime has an impact on peak consumption, Faruqui et al. (2017)

found that requiring consumers to opt-out rather than opt-in to having time varying electricity prices

led to a further 3.9% reduction in consumption in the peak demand period.

4.5 What this means

International studies of short run responsiveness to price changes confirms the result of the

review of the long run responsiveness – that is, contrary to concerns, households are responsive to

substantial price differences. Technology assists in generating a greater response to price incentives,

particularly if responses are automated, but is not necessarily required for prices to have an effect.

The limited evidence suggests that the simple change of making participation in cost reflective

pricing opt-out rather than opt-in will considerably increase its uptake.

The results from the studies of Ito (2014), Byrne et al. (2019) and Deryugina et al.

(forthcoming), in combination with those for shorter time periods, are suggestive of the conditions

under which households are most likely to substantially respond to the incentives provided by cost

reflective pricing. The conditions under which moderating consumption will result in a financial

payoff need to be clear to the household. Reducing consumption on specified days or for specified

hours is easier than to avoid breaching a threshold that is not observed. The changes need to be

permanent and greater changes will be observed over time as it takes time and possibly further

investments to alter electricity consumption. A reduction in demand during short periods of time,

rather than over a whole month, may not be quantitatively large but could still be economically

important in reducing the need for costly infrastructure. An analysis of demand data for Victoria

from the 1/11/2016 to 31/10/2017 found the peak demand was over 8400MWs. But demand only

exceeded 7000MWs on eighteen days. If demand could be moderated on these days back towards

21
Automation was not found to have any effect on uptake of real time pricing or dynamic tariffs.

22
7000MWs considerably less capacity would be required. Lower capacity means less costs to be

recovered which ultimately means lower prices.

The limitation of all this work is that it doesn’t inform on the second major concern around

the equity impacts of cost reflective pricing. We review the limited evidence on this in the next

section.

5 Price responsiveness by different types of households

The second of the concerns for cost reflective pricing is the potential impact it could have on

lower income households. If low income households are less price responsive22, their bills could go

up and if they are more price-responsive, they may be exposing themselves to a greater risk of, for

example, health problems due to heat. There is, however, little systematic evidence on the effect of

cost reflective pricing for lower income households, with inconsistent findings across the limited

research. Researchers have attempted to generate evidence by directly measuring effects in surveys,

analysing how consumers report about different pricing plans and how the bills of consumers with

different incomes would change under different pricing plans.

5.1 Responsiveness by household income

The literature on responsiveness by household income level is somewhat mixed. In 2012,

Frontier Economics and Sustainability First undertook a literature review of 30 residential customer

demand side response trials for the UK Department of Energy and Climate Change. The authors

found vulnerable and low-income consumer responsiveness to demand side response initiatives

varies across studies, with some finding that these groups respond less than the average response.

There was little evidence on the impact of demand side response incentives on low-income and

22
An implicit assumption underlying this concern is that lower income households could be less price
responsive than other household and would wear any price increase at a cost to other items in an already
stretched budget. Some reasons for being less price-responsive could be that lower income households are:
 already mindful of their energy consumption and are limited in making further changes
 generally in less energy efficient homes
 have older and less energy efficient appliances
 are renting and cannot make physical changes to the property;
 have more health needs which require better thermal comfort such as those with children or
the elderly.

23
vulnerable consumers in the UK in particular. In the US, Deryugina et al. (forthcoming) found little

evidence that elasticities depended on economic characteristics, such as income and education, but

did find some variation across age and race.

Faruqui et al. (2010) found mixed results from reviewing five programs for differences in

price responsiveness between low and high income households. Elasticities were referred to but not

reported in the paper, but two of the programs found low income customers were equally

responsive to average customers, while the other three studies found that they were less responsive

(but responsive nonetheless).

5.2 Examination of bills for impacts of cost reflective pricing

An alternative way to analyse household impact is to analyse existing bills to see how much

lower income households could be impacted by a move to cost reflective pricing. The limitation is

that current consumption by a household does not reflect any behaviour change in response to a

different price structure.

Simshauser and Downer (2016) undertook an analysis of smart meter bills from Victoria

merged with information from AGL Energy’s household survey to analyse what types of customers

might gain from a shift from flat rate to a cost reflective type tariff (time of use plus critical peak

pricing). They consider six types, two of which include groups that are typically characterised as low

income and/or vulnerable: “Households in Hardship” and “Concession and Pensioner”. When no

allowance for behaviour change was made, the “Households in Hardship” group had the greatest

share of households that gained (65%) because their daily consumption profile was such that they

gained from off-peak discounts without losing too much in the peak. Only 45% of the “Concession

and Pensioner” group gained though. Once some demand responsiveness was included in the

calculations the majority of these groups (79% and 62%) gained – indeed 64% of all consumers

24
gained related to a flat rate system. This assumes overall revenue doesn’t change. If cost reflective

pricing reduces peak demand, costs and revenue will fall and more households will gain.23

5.3 Low income households in Australia

In Australia, there is no direct evidence on the impact of cost reflective pricing on low

income households similar to the studies above, but there are studies which examine household

sentiment and potential behaviour if there were cost reflective pricing. One study asked consumers

how they felt about different pricing plans with analysis based on income group (Stenner et al.,

2015). The authors found that critical peak pricing had the greatest, albeit small, appeal to lower

income consumers which could reflect the desire for cheaper electricity for most of the year and

avoidance of usage at critical times. In addition, flat rates were increasingly preferred over critical

peak pricing as household income rises. On the other hand, Nicolls et al. (2015) asked vulnerable

households about cost reflective pricing and found considerable concern about it.

5.4 Potential relationship between income and types of appliances

To the extent electricity consumption is driven by appliances, the use of energy consuming

appliances might be expected to increase with income. However, it is also possible that higher

income households can afford more efficient appliances.

In Sydney and regions, Fan et al. (2015) examined a range of factors and found that a pool

pump drives the most consumption, with increased daily consumption by about 50 per cent on an

average annual basis when compared to those who did not have a pool pump. The next feature that

drove consumption was air conditioning. Gas hot water was found to decrease daily electricity

consumption by 18% on an average annual basis24. Owning a pool pump and the extent of air

conditioning is likely to be positively correlated with income which suggests that households that

23
These results assume a certain pricing pattern and consumer demand profiles. Horowitz and Lave (2014) do
a similar exercise (without demand response) for households in northern Illinois and Chicago and find only 25
per cent of customers would be better off and only 19 per cent of low income customers.
24
This data includes half hourly demand for 12 months from households in the Sydney region, as well as a
range of other variables.

25
consume more electricity are more likely to save the most money, if the findings of Horowitz and

Lave (2014) applied in Australia.

5.5 What this means

The limited mixed evidence of low income and vulnerable households’ responsiveness is

consistent with cost reflective pricing not automatically having a uniform impact. The surveys

undertaken in Australia on the perceptions of low income households indicate that some households

are apprehensive about the potential impact, while other households perceive they may be able

take advantage of cost reflective pricing to make savings. An analysis of bills would be a useful

starting point to better understand potential impacts. Concessional cost reflective pricing could also

be considered. For instance, a cost reflective pricing scheme could feature discounts for concession

holders. So although concession householders would pay a differential price for peak and off-peak

times, both the peak and off-peak prices could be well below the prices paid by non-concession

holders.

6 Discussion and conclusion

With technological change reducing the costs of renewable and distributed generation, and

with the prospect of a substantial shift in demand due to the uptake of electric vehicles,

considerable investment in new infrastructure is likely. Hence it is even more important to have a

pricing structure that encourages efficient use and investment of electricity infrastructure, which will

be better than the prevailing flat rate structure. Governments have been reluctant to proceed too

far with pricing reform though for two reasons. First, there is an active public debate around

whether prices can have a significant influence on consumer choices. Second, there is a concern that

moving away from flat rates for electricity could exacerbate energy affordability issues for the lowest

income households.

The empirical literature on the importance of each of these conditions provides substantial

international and local evidence that households, in the long run, respond to electricity prices. While

the demand for electricity is found to be inelastic, households do respond in a material way to an

26
increase in price. And this responsiveness is greater than what is estimated for some other

necessities, like petrol and water. The responsiveness of the demand for electricity is comparable to

that for cigarettes and the greatest estimates for electricity are similar to those for milk and bread.

There is also a substantial body of international evidence from pilots and trials that

consumers respond to price signals that encourage moderating consumption during peak periods.

While these very short run price elasticities are smaller than the long run estimates, this is consistent

with economic theory that people can adapt over longer time frames by making more efficient

investment decisions. Furthermore, the very short run responsiveness is increased if pricing reform

is accompanied by technology that improves the information to households and eases their ability to

respond to price changes. Pricing reform is also found to be more likely to work if households must

opt-out, rather than opt-in, to cost reflective pricing.

There are caveats on applying elasticities compiled from pilots and experiments to an entire

population as pilots and experiments can involve self-selection and incentives to participate which

may not hold in other circumstances. However, the weight of evidence would support that

households are responsive to price changes through adjusting demand. Further pricing reform is

likely to be effective and at the very least, further local research on the impacts of pricing reform

should be undertaken.

There is insufficient evidence to determine whether low income and vulnerable households

respond less to electricity prices compared to other households. There is no Australian quantitative

evidence and the international evidence is ambiguous on the direction and size of any effects. The

qualitative evidence seeking to understand whether low income and vulnerable households are

more likely to be adversely affected by electricity price reforms is mixed. Future Australian research

could improve focus on more precisely understanding the impact of cost reflective electricity prices

on households by including a sample of a sufficient size of low income and vulnerable Australians.

This would support reform efforts. Estimating the responsiveness of demand changes over time

based on a panel of households or communities for a substantial period of time is a recommended

27
area for this research. If, for example, it turns out that cost reflective pricing benefits many low

income earners (through providing savings most of the year that offset higher prices in the peak

periods) then this would bolster the cases for making cost reflective pricing opt-out rather than opt-

in and for making the default offer a cost reflective rather than flat rate.

In addition, there could be complementary policies to support a change to cost reflective

pricing for low income and vulnerable households, such as including discounts for concession

holders, providing additional information to concession households to support their decision-making

to opt-in or not to the new pricing system, once-off discounts for new appliances or a targeted

program of energy efficiency retrofits for homes e.g. for social housing tenants. The likely

effectiveness of relevant policies that would complement the change to cost reflective pricing would

need to be examined further. Access to data for researchers appears to be a key limitation. Most

short run pricing studies appear to partner with energy companies to access customer data. In

addition, similar to other research areas, details about income can sometimes be difficult to access

as disclosure through surveys is limited. In Victoria, smart meters are in place which means

household usage data exists and is held by electricity companies. Access to this data requires

working with electricity companies with permission from households. The literature for Australia,

and indeed Victoria, is limited and further studies would assist in better understanding the dynamics

of household electricity use.

References

ACCC (2018), Restoring electricity affordability & Australia’s competitive advantage, Final report on

the inquiry into retail electricity supply and pricing.

AEMO (2019), 2019 Electricity Statement of Opportunities, August 2019.

Allcott, Hunt (2011), “Rethinking real-time pricing”, Resource and Energy Economics, 33, 820-842.

Ausgrid (2015), “Tariff structure statement – Appendix 5: Price elasticity of demand”, found on the

Australian Energy Regulator website at https://www.aer.gov.au/networks-

28
 pipelines/determinations-access-arrangements/pricing-proposals-tariffs/ausgrid-tariff-

structure-statement-2017/proposal.

Ben-David, Ron (2018), “Competition, neo-paternalism and the Nonsumer uprising”, paper

presented at the Consumer Law Conference, Sydney (15 May 2018), found at

https://www.esc.vic.gov.au/media-centre/competition-neo-paternalism-and-nonsumer-

uprising

Borenstein, Severin (2013), “Effective and equitable adoption of opt-in residential dynamic

electricity pricing”, Review of Industrial Organization, 42, 127-160.

Breunig, Robert and Carol Gisz (2009), “An exploration of Australian petrol demand: Unobservable

habits, irreversibility and some updated estimates”, Economic Record, 85(268), 73-91.

Byrne, David, Andrea La Nauze and Leslie A. Martin (2019), “An experimental study of monthly

electricity demand (in)elasticity”, (May 17, 2019). Available at SSRN:

https://ssrn.com/abstract=3164697

Deryugina, Tatyana, Alexander MacKay and Julian Reif (forthcoming), “The long-run dynamics of

electricity demand: Evidence from Municipal Aggregation”, American Economic Journal:

Applied Economics.

Doojav, Gan-Ochir and Kaliappa Kalirajan (2019), “Income and price elasticities of electricity demand

in Australia: Evidence of state-specific heterogeneity”, Australian Economic Papers, 58, 194-

206.

Fan, H., I.F. MacGill, A.B. Sproule (2015), “Statistical analysis of driving factors of residential energy

demand in the greater Sydney region, Australia”, Energy and Buildings, 105, 9-25.

Fan, Shu and Rob Hyndman (2011), “The price elasticity of electricity demand in South Australia”,

Energy Policy, 39, 3709-3719.

Faruqui, Ahmad, Sanem Sergici and Jennifer Palmer (2010), “The impact of dynamic pricing on low

income customers”, Institute for Electric Efficiency (IEE) White paper.

29
Faruqui, Ahmad, Sanem Sergici and Lamine Akaba (2014), “The impact of dynamic pricing on

residential and small commercial and industrial usage: New experimental evidence from

Connecticut”, The Energy Journal, 35(1), 137-160.

Faruqui, Ahmad, Sanem Sergici and Cody Warner (2017), “Arcturus 2.0: A meta-analysis of time-

varying rates for electricity”, The Electricity Journal, 30(10), 64-72.

Finkel, Alan and Australia, Department of Environment and Energy, issuing body (2017), Independent

Review into the Future Security of the National Electricity Market. Australia, Department of

Environment and Energy, June 2017.

Frontier Economics and Sustainability First (2012), Demand side response in the domestic sector – a

literature review of major trials, August, 2012, available at

https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_

data/file/48552/5756-demand-side-response-in-the-domestic-sector-a-lit.pdf, accessed

11/9/2019.

Gans, Joshua, Stephen King and N. Gregory Mankiw (1999), Principles of microeconomics, Harcourt

Australia.

Gillan, James M. (2018), “Dynamic pricing, attention and automation: Evidence from a field

experiment in electricity consumption”, mimeo

Grafton, R. Quentin and Michael B. Ward (2008), “Prices verus rationing: Marshallian surplus and

mandatory water restrictions”, Economic Record, 84, S57-S65.

Hasan, Syed and Mathias Sinning (2018), “GST Reform in Australia: Implications of Estimating Price

Elasticities of Demand for Food”, Economic Record, 94(306), 239-254.

Horowitz, Shira and Lester Lave (2014), “Equity in residential electricity pricing”, The Energy Journal,

35(2), 1-23.

Infrastructure Victoria (2018), Advice on Automated and Zero Emissions Vehicles Infrastructure,

October, 2018.

30
Ito, Koichiro (2014), “Do consumers respond to marginal or average price? Evidence from non-linear

electricity pricing”, American Economic Review, 104(2), 537-563.

Jemena (2019), Jemena Power Changers trial evaluation report. Available from

https://www.energy.vic.gov.au/electricity/demand-management-achievements (accessed

July 1st, 2019)

Kim, Hyungkwan, Paul V. Preckel, James S. Eales, Douglas Gotham and Andrew L. Liu (2015), “Impact

Assessment of Time of Use Pricing for Electricity: Evidence from a Natural Experiment in

Ontario”, selected paper prepared for presentation at the 2015 Agricultural & Applied

Economics Association and Western Agricultural Economics Association Annual Meeting San

Francisco, July 2015.

Labandeira, Xavier, José M. Labeaga and Xiral López-Otero (2017), “A meta-analysis on the price

elasticity of energy demand”, Energy Policy, 102, 549-568.

Motlagh, Omid, Greg Foliente and George Grovez (2015), “Knowledge-mining the Australian Smart

Grid Smart City data: A statistical-neural approach to demand-response analysis”, Chapter

10 in S. Geertman et al. (eds.), Planning support systems and Smart Cities, Springer

International Publishing Switzerland.

Narayan, Paresh Kumar and Russell Smyth (2005), “The residential demand for electricity in

Australia: an application of the bounds testing approach to cointegration”, Energy Policy, 33,

467-474.

Newsham, Guy R. and Brent K. Bowker (2010), “The effect of utility time-varying pricing and load

control strategies on residential summer peak electricity use: A review”, Energy Policy, 38,

3289-3296.

National Institute of Economic and Industry Research (2007), The own price elasticity of demand for

electricity in NEM regions, a report prepared for the National Electricity Market

Management Company.

31
Nicholson, Moira L., Michael J. Fell and Gesche M. Huebner (2018), “Consumer demand for time of

use electricity tariffs: A systematized review of the empirical evidence”, Renewable and

Sustainable Energy Reviews, 97, 276-289.

Nicolls, L., McCann H., Strengers Y. and Bosomworth K. (2017), “Heatwaves, Homes & Health: Why

household vulnerability to extremes is an electricity policy issue”, Centre for Urban

Research, RMIT University, Melbourne.

Pham, V. and D. Prentice (2013) “A random coefficients logit analysis of the counterfactual: A merger

and divestiture in the Australian cigarette industry”, La Trobe University School of Economics

Working Paper No.4 (for 2013).

Rai, Alan M., Luke Reedman and Paul Graham (2014), “Price and income elasticities of residential

electricity demand: the Australian evidence”. Proceedings of the Australian Conference of

Economists, 2014.

Simshauser, Paul and David Downer (2016), “On the inequity of flat-rate electricity tariffs”, The

Energy Journal, 37(3), 199 – 229.

Stenner, Karen, Elisha Frederiks, Elizabeth V. Hobman and Sarah Meikle (2015), Australian

consumers’ likely response to cost-reflective electricity pricing, CSIRO Energy Flagship.

Wood, Tony, Lucy Carter and Cameron Harrison (2014),”Fair pricing for power”, Grattan Institute.

32
Appendix 1

Infrastructure Victoria recommendations

Mandate cost reflective pricing for all energy customers within 0-5 years to fully realise the

benefits of smart meters, increase customer engagement on energy consumption patterns and

influence customer choices to reduce peak and potentially total energy demand. The first step to

achieving this would be to provide leadership in working with industry to increase customer

awareness on the benefits of cost reflective tariffs, benefits that include fairer pricing and lower

costs pricing.

ACCC recommendations

The ACCC considers that steps should be taken to accelerate the take up of cost reflective

network pricing.

Governments should agree to mandatory assignment of cost reflective pricing on retailers,

ending existing opt-in and opt-out arrangements.

Mandatory assignment of the network tariff should apply for all customers of a retailer that

have metering capable of supporting cost reflective tariffs (that is, a smart of interval meter).

Retailers should not be obligated to reflect the cost reflective network tariff structure in

their customers’ retail tariffs, but should be free to innovate in the packaging of the network tariff as

part of their retail offer.

Given the potential for negative bill shock outcomes from any transition to cost reflective

network tariffs should retailers pass these network tariffs through to customers, governments

should legislate to ensure transitional assistance is provided for residential and small business

customers. This assistance should focus on maximising the benefits, and reducing the transitional

risks, of the move to cost reflective pricing structures. This includes:

 A compulsory ‘data sampling period’ for consumers following installation of a smart

meter

 A requirement for retailers to provide a retail offer using a flat rate structure

33
  Additional targeted assistance for vulnerable consumers.

Demand tariffs, which charge retailers based on their customers’ maximum demand during

pre-determined typical system peak times, represent an appropriate structure for the initial

mandatorily assigned network tariffs. This tariff structure provides a balance of the objectives of cost

reflectivity, simplicity and price certainty.

We note that the extent to which cost reflective tariffs can be introduced is limited to the

extent that a retailer’s customers have smart (or interval) meters. We therefore note the importance

of the ACCC’s recommendation 15, on supporting to uptake of smart meters, in achieving outcomes

in this area.

Governments should appropriately fund communication campaigns around the benefits of

cost reflective pricing and smart meters to build community acceptance and awareness of individual

and community wide benefits, as well as customer awareness of their rights.

34
Appendix 2 - Elasticities, electricity and equity

Elasticities are an important concept in economics. They tell us about how responsive

people are to price through quantitatively summarising how much people change their consumption

in response to a change in price.25 After reviewing the economics of elasticities, we analyse the

determinants of the elasticity of demand for electricity, followed by a discussion of how elasticities

can provide useful information for dealing with negative equity consequences.

Demand curves

In general, people will consume more of a good if the price is low. If the price is higher, then

people will consume less. This is the law of demand and is represented by a downward sloping

demand curve (see Figure 1). There are also factors which can shift a demand curve (also see Figure

1). These are known as exogenous factors, such as weather or a change in income. For instance, a

household could choose to consume more electricity at the same price due to a hot day, or use more

appliances because their household income has increased. The demand curve could also shift due to

the purchase of new appliances, such as an air conditioner or pool pump. This shift could also be a

decrease in demand from the grid though the purchase of solar panels or the adoption of more

energy efficient energy appliances.

Elasticities and the shape of demand curves

Elasticity is a measure of how much buyers and sellers respond to change in market

conditions and allows more precise analysis of supply and demand (Gans et al., 1999)26. The price

elasticity of demand is the percentage change in quantity demanded in response to a one percent

change in price. For instance, an elasticity of -0.2 means that the quantity demanded falls by 0.2 per

cent in response to a one per cent increase in price. This elasticity can differ along the demand

25
This paper concentrates on how the demand for electricity responds to a change in the price of electricity.
Electricity consumption will also respond to the price of other goods, income and the weather. It is
straightforward to define and possible to estimate elasticities of demand with respect to these products,
interpretable in the same way (see Gans et al. (1999) for more details).
26
This is a contemporary introductory microeconomics text book.

35
curve.27 One of the main advantages of an elasticity is that as it is a proportional change, it is unit-

free which enables comparison of the results of different studies considering different changes in

prices.

Figure 1: Demand curve and shifts

The slope of the demand curve can represent how responsive buyers are to a change in

price. A demand curve that is steep is said to be ‘inelastic’ which means that for a given price

change, the quantity demanded changes less than proportionally. A demand curve that has a less

steep gradient is said to be ‘elastic’ which means that for a given change in price, the quantity

demanded will change by a greater proportion. Buyers are more sensitive, or responsive, to price

changes than when the demand curve is more inelastic. It is only when demand for a good is

perfectly inelastic that buyers are unresponsive to price28. Figure 2 represents these shapes for

demand curves.

27
There is also a link between the elasticity of demand and household expenditure. Household expenditure is
price times the quantity. If demand is relatively inelastic, then changes in price cause total expenditure to
move in the same direction – that is, with inelastic demand if the price goes up, then the household
expenditure on that good will also go up. This may be of concern to households with relatively low incomes.
28
A demand curve can also be perfectly elastic – any increase in the price of the good will cause demand to
drop to zero.

36
Figure 2: Shape of demand curves and responsiveness to price

Elastic demand (E>1): Q changes proportionally Inelastic demand (E<1): Q changes

more than P proportionally less than P

Perfectly inelastic (E=0): P changes, but Q remains the same

Broad determinants of demand elasticities

There are some broad determinants of how elastic demand may be. This include whether a

good is a necessity, the time horizon under consideration and the type of the consumer.

If a good is a necessity, then the quantity demanded by a buyer tends to not change by

much if there is a price change. This means that demand for necessity goods tends to be relatively

37
more inelastic compared to goods which are discretionary purchases. For electricity, we would

expect demand to be relatively inelastic as it is something we use every day.

The time horizon for a good can influence the nature of the elasticity. This issue is

particularly acute for electricity as it is almost exclusively consumed upon purchase. Consumers are

able to adjust to a different extent over different time periods. Adjustment at a point in time may

involve substituting consumption over time, such as doing the laundry or charging an electric car in

an off-peak period. Long run elasticities tend to be more elastic than short run elasticities. For

instance, a person may not be able to change their electricity consumption today by much (as, for

example, they need their car for the next day), but over time they can buy more energy efficient

appliances which can result in lower energy consumption over a longer time period. Electricity is a

complementary good to several significant products like air-conditioning, pool pumps, ovens and

fridges. A rise in the electricity price will also affect the demand for these goods. Indeed, it is likely

that the long run effects of a change in the price of electricity would result in consumers adjusting

their set of appliances over time (for instance, replacing air conditioners or adjusting processes to

reduce the electricity bill). Forming habits and learning about consumption patterns can also

increase price responsiveness over a longer time period.

Price elasticities are also likely to be greater if a price increase is permanent rather than

temporary. If a price increase is known to be temporary a household is less likely to make changes to

its behaviour, let alone to appliances or insulation, than if a price change is known to be permanent.

Lastly, there can be differences in elasticities depending on the type of consumer. For

instance, some industrial consumers may have more substitutes for electricity available, such as gas,

or substituting times when they can consume electricity, such as choosing to run components of

their plants during off-peak pricing times. By comparison, a household may have less choice of when

it uses electricity. This suggests that industrial consumers tend to be more price elastic than

households. The same could be said for cohorts within households - lower income households could

38
respond differently to higher income households. It could be that higher income households over

time can invest in energy efficient appliances compared to lower income households.

39